automake-cli 0.4.8

The AI-native shell that turns natural language into actions.

🤖 automake: Your AI Command-Line Assistant

The AI-native shell that turns natural language into actions.

---

---

automake is a Python-based, agent-first command-line tool that uses a powerful multi-agent system to interpret your natural language commands and execute them.

Forget remembering complex flags or exact Makefile targets. Just tell automake what you want to do.

✨ Key Features

• AI-Native Commands: Run terminal commands, execute Makefile targets, and perform other tasks using plain English.
• Multi-Agent System: A sophisticated ManagerAgent orchestrates a team of specialists (Terminal, Coding, Web, Makefile) to get the job done right.
• Interactive Agent Mode: Launch a chat session with automake agent for complex, multi-step tasks.
• Local First: Integrates with local LLMs via Ollama for privacy and offline access.
• Intelligent Error Handling: If you make a typo, the agent will analyze the error and suggest a correction.
• User-Friendly & Configurable: A clean CLI and a simple config.toml file for all your settings.

⚙️ How It Works

automake puts an AI agent at the heart of your command line, following a modern, agentic workflow:

1. Parse Prompt: The CLI captures your natural language instruction (e.g., automake "list all python files").
2. Invoke Agent: Your prompt is sent directly to a central Manager Agent.
3. Reason & Delegate: The agent analyzes your request and decides which specialist is best for the job—the TerminalAgent for shell commands, the MakefileAgent for make targets, etc.
4. Execute & Observe: The specialist agent executes the task, and the result is observed.
5. Stream Output: The results are streamed directly to your terminal in real-time.

This entire workflow is triggered by the run command. For example: automake run "list all python files".

🚀 Getting Started

Prerequisites

• Python 3.13+
• uv (recommended for installation)
• An active Ollama server with a running model (e.g., ollama run qwen3:0.6b).

Installation

There are several ways to install and run automake.

Recommended: uv tool install

For a permanent installation, we recommend using uv tool install. This makes the automake command available globally in a dedicated environment.

1. From Git (Bleeding Edge) Install the very latest version directly from this repository:

# Install the latest version from the main branch
uv tool install git+https://github.com/biokraft/auto-make.git

# You can also install a specific tag or branch
uv tool install git+https://github.com/biokraft/auto-make.git@v0.4.8

2. From PyPI (Stable) Install the latest stable release from PyPI:

uv tool install automake-cli

Temporary Execution: uvx

If you prefer not to install the tool, you can run it directly using uvx (similar to npx). This downloads and runs the package in a temporary environment.

uvx automake-cli run "your command here"

Traditional pip

You can also use pip for a standard installation:

pip install automake-cli

First-Time Setup

After installation, run the initialization command once to set up Ollama and download the required model:

automake init

This command will:

• Verify that Ollama is installed and running.
• Download the configured LLM model if not already available.
• Ensure everything is ready for natural language command interpretation.

✍️ Usage

Non-Interactive Commands

The primary way to use automake is with the run command, passing your request as a string argument:

# Run a terminal command
automake run "recursively find all files named README.md"

# Execute a Makefile target
automake run "run the tests and generate a coverage report"

# Build the project
automake run "build the project"

Interactive Agent Session

For more complex tasks, start a chat with the agent:

automake agent

For detailed usage information and available options, run:

automake help

🛠️ Configuration

automake features a modern, user-friendly configuration system. On first run, it creates a config.toml file in your user configuration directory.

Setting the AI Model

Configure your preferred Ollama model:

# Set a specific model
automake config set ollama.model "qwen2.5:7b"

# After changing the model, initialize it
automake init

Important: After changing the model, you must run automake init to download and initialize the new model if it's not already available locally.

View and Modify Configuration

• View current config: automake config show
• Set specific values: automake config set <section.key> <value>
• Edit manually: automake config edit
• Reset to defaults: automake config reset

Common Configuration Examples

# Set the AI model
automake config set ollama.model "qwen3:8b"

# Set the Ollama server URL
automake config set ollama.base_url "http://localhost:11434"

# Set logging level
automake config set logging.level "DEBUG"

# Set AI interaction threshold
automake config set ai.interactive_threshold 90

Configuration Structure

Run automake config show to see the current configuration.

# Example from automake config show
[ollama]
base_url = "http://localhost:11434"
model = "qwen3:0.6b"

[logging]
level = "INFO"

[ai]
interactive_threshold = 80

🎬 Demos

Want to see some UI/UX demos? Just run uv run make demo-all or use automake: automake run "show all demos"

Note: Running demos with automake may cause animation display issues. For the best demo experience, use the direct uv run make demo-all command.

🗺️ Project Roadmap

For a detailed breakdown of the project roadmap, implementation phases, and technical specifications, see SPECS.md.

📜 Changelog

All notable changes to this project are documented in the CHANGELOG.md file.

📄 License

This project is licensed under the Apache 2.0 License. See the LICENSE file for details.